Acetylcholine applied extraluminally to isolated, perfused dog mesenteric artery segments produced an endothelium-dependent depressor response when the perfusion pressure was raised by continuous infusion of noradrenaline; the potency was 1/30 to 1/60 that of intraluminal acetylcholine. Contractions induced by transmural electrical stimulation were attenuated by treatment with intra- and extraluminal acetylcholine; the inhibitory effect of intraluminal acetylcholine was greater than that of extraluminal acetylcholine. Removal of endothelium did not significantly alter the inhibitory effect. In mesenteric artery strips with endothelium, treatment with oxyhaemoglobin suppressed the relaxant response to acetylcholine but did not influence the inhibitory effect of acetylcholine on stimulation-evoked contractions. Acetylcholine reduced the 3H-overflow and contraction of superfused mesenteric artery strips, preloaded with 3H-noradrenaline, response to transmural stimulation. By the use of bioassay (dog femoral artery segment with endothelium/coronary artery strip without endothelium), the release of EDRF was first determined in the perfusate, which was introduced to dog mesenteric artery strips loaded with 3H-noradrenaline. The 3H-overflow and contraction caused by the stimulation were not attenuated by EDRF and were also observed following treatment with superoxide dismutase. Inability of the perfusate to reduce the stimulation-evoked 3H-overflow was also observed when the donor and assay tissues were treated with superoxide dismutase. It may be concluded that the inhibition by acetylcholine of the release of neuronal noradrenaline is not dependent on endothelium. Extraluminally applied acetylcholine would reach the endothelium and release EDRF, and intraluminal acetylcholine is presumed to act directly on prejunctional muscarinic receptors; however, acetylcholine appears to cross the medial layer more efficiently from intima to adventitia than in the reverse direction.